Risk of clinical mastitis within lactation in Dutch dairy cattle

The number of clinical mastitis cases dairy cows may experience over a lifetime has been a subject of several studies. Some of them analysed repeated events by different types of statistical models and used different approaches such as retaining only cases after certain time or considering only the first case within lactation. No studies included the two types of heterogeneity that is often observed in field. One derived from lack of independence of individuals from the same herd and the second, from intra-subject correlation of repeat events. Our objectives were to estimate the risk of clinical mastitis within-lactation for Dutch dairy cattle accounting for conditional dependence between cases and to assess risk factors associated with a changing risk. Data were obtained from farmers' records of 248 farms participating in a voluntary programme. Data included all lactations present during 01/1999 to 04/ 2004 (n=33,105) of 22,603 cows. Data were analysed with a survival analysis conditional model for repeated events using a sojourn time-scale between cases corresponding to time since last event and a frailty term. Higher hazard rates are positively associated with parity number, length of lactation, 305-days milk production, total number of clinical cases in the herd, year of calving and the interaction herd size-305-days milk production. Herd size and the rest of the interactions were negatively associated. However, although number of days open and the lactation value should be incorporated, their effect is not significant. Our estimations serve as guidelines to farmers for improving udder health. In addition, our results and the methodology employed are useful for exploring other aspects of associations between genetic resistance of cattle and occurrence of clinical mastitis.